A history of Time Aligned Loudspeakers

[kimmosto]

Something wrong with ^^ the Y axis is in Pascals (sound pressure) - once the driver has reached the end of the step, there is no further motion, so the SPL decays.

That is theoretical ideal step response visualizing effect of crossover only. Not driver or anything with limited band and high pass slope. Pa unit is there because graph is generic impulse response chart and response data was taken from SPL chart.

 

[Thomas_A]

Just some reminders.

Step response is not a good way to visualise time or phase alignment in regard of audibility, group delay should be used.

The typical electrical 6 dB/octave crossovers are usually not sufficient, the total acoustic slope which counts for the filter characteristic is usually higher so it was a bit of a misunderstood myth.

What makes 6dB filters so special – Just Blogging

Is it possible to make a first-order crossover in a real-world situation? – Just Blogging

Also such a shallow slope has also significant disadvantages on the vertical radiation unless a coincident design is used, even the "time incident step" won't be such anymore at vertical angles and thus for example from reflections from the floor and the ceiling.

I saw from a comment from the blog "The only company, who really used the true 1st order crossover is no longer in business. "

Is that true? Or is it just so that he do not know every company?

 

[thewas]

I saw from a comment from the blog "The only company, who really used the true 1st order crossover is no longer in business. "

Is that true? Or is it just so that he do not know every company?

I don't think KHF (like anyone else) knows every company, do you have some concrete companies in mind?

 

[Thomas_A]

I don't think KHF (like anyone else) knows every company, do you have some concrete companies in mind?

I was just thinking of Ino Audio that has been around since 1980 or so. It is more of an underground company so few know it. But the constructor (Ingvar Oehman) does not really point any importance of audibility, which is quite unusual for a loudspeaker company: I cite Ingvar from an older loudspeaker catalogue translated to English:

One of the features of the pi60 requires comment – the phase linearity
The Ino pi60 is a phase linear and thus waveform reproducing system. As a manufacturer of a product, you are of course supposed to argue that all the features (especially the unusual ones) that your products possess are extremely important, but I, with my background in scientific research, do not want to argue things that are not true.
I consider the phase linearity of the pi60 family as a bonus. An unusual feature and not at all insignificant, but not nearly as important as all the other qualities of the speakers. Almost none of the speakers on the market are phase linear, but most of them (or rather, listeners to them) suffer from many much bigger problems than the fact that the speakers are not phase linear! ☺
A loudspeaker's overall phase characteristics (phase linearity, phase integration between the elements, phase integration in the horizontal plane, group delay, etc.) largely determine how it is perceived, but phase linearity is much less important than most people think. Especially if you put it in relation to things like tone curve linearity, freedom from distortion, phase homogeneity, dispersion characteristics and room adaptation.
The dominant faults of most loudspeakers are rather high distortion and a crooked tone curve (seen from a psychoacoustic point of view, of course).

 

[KL....]

I don't think KHF (like anyone else) knows every company, do you have some concrete companies in mind?

Thank you, @thewas, had a chuckle (to myself) but 20 or so years ago there were some companys useing Concrete and/or Concrete/Marble enclosures, such as Norh....

 

[Thomas_A]

Thank you, @thewas, had a chuckle (to myself) but 20 or so years ago there were some companys useing Concrete and/or Concrete/Marble enclosures, such as Norh....

Regarding history, there are companies with concrete speakers today. E.g. Rauna, founded 44 years ago:

www.rauna.com

 

[thewas]

Thank you, @thewas, had a chuckle (to myself) but 20 or so years ago there were some companys useing Concrete and/or Concrete/Marble enclosures, such as Norh....

Yes, its always weird and funny to use that word in English.

 

[Thomas_A]

Yes, its always weird and funny to use that word in English.

I somehow avoid the English concrete, in the meaning specific. I feel like I am talking "Swenglish" since the same word in Swedish is "konkret". Luckily the matter "concrete" is "betong" in Swedish.

 

[KL....]

Just some reminders.

Step response is not a good way to visualise time or phase alignment in regard of audibility, group delay should be used.

The typical electrical 6 dB/octave crossovers are usually not sufficient, the total acoustic slope which counts for the filter characteristic is usually higher so it was a bit of a misunderstood myth.

What makes 6dB filters so special – Just Blogging

Is it possible to make a first-order crossover in a real-world situation? – Just Blogging

Also such a shallow slope has also significant disadvantages on the vertical radiation unless a coincident design is used, even the "time incident step" won't be such anymore at vertical angles and thus for example from reflections from the floor and the ceiling.

Thank you, @thewas, my speakers (3 driver/2.5 way designed/built (M/T/MW arrangement, 150mm/32mm/165mm), all Drivers in-phase, min Impedance is 6/6.5ohms, SPL 91/92dB, F3/5 is 25Hz (Vented) to 26kHz, nearly 30yrs ago) utilise 1st Order 6 dB/octave XO (in conjunction with Driver rolloff where appropriate) but these required very careful matching/selection of the Drivers and assist of the 32mm (outward) Dome Tweeter Dispersion characteristics (XO is 1.1/1.2kHz) so that at the XO point (and a little higher) it performed more like a Mid Driver. Not perfect, 3/4dB rise/bump at 28Hz then 32-22kHz is +/- 1.5dB, especially as the variety of Drivers then was dramatically less then what is now available, but works well and now time for a Mid Driver update (weakest performing Driver). This suggests that it is possible.... although power handleing is about 100W.

 

[SportyQ]

These speakers are excellent examples of misplaced priorities in audio. If you stagger your speakers on a baffle like that, you will ruin vertical directivity and possibly horizontal directivity as well, if not done properly. Bear in mind, sound travels 343mm in 1ms. So if the baffle is stepped by 34mm, it will provide 0.1ms of time alignment. Which do you think is more audible? A 0.1ms time alignment, or poor vertical directivity?

Are you talking about the D'Apolito alignment ? It was likely done on purpose for exact complaint you make. To get cancelations off axis and especially at room boundaries - vertically. IIRC.
My personal time align project was built in the style of Thiel and Sica, with sloped baffle

 

[KL....]

And thank you for your attention!

Thank you, @Trifonov Audio, what would be consideration/thoughts of the following....

.... my speakers (3 driver/2.5 way designed/built (M/T/MW arrangement, 150mm/32mm/165mm), all Drivers in-phase, min Impedance is 6/6.5ohms, SPL 91/92dB, F3/5 is 25Hz (Vented) to 26kHz, nearly 30yrs ago) utilise 1st Order 6 dB/octave XO (in conjunction with Driver rolloff where appropriate) but these required very careful matching/selection of the Drivers and assist of the 32mm (outward) Dome Tweeter Dispersion characteristics (XO is 1.1/1.2kHz) so that at the XO point (and a little higher) it performed more like a Mid Driver. Not perfect, 3/4dB rise/bump at 28Hz then 32-22kHz is +/- 1.5dB, especially as the variety of Drivers then was dramatically less then what is now available, but works well and now time for a Mid Driver update (weakest performing Driver). This suggests that it is possible.... although power handleing is about 100W.

 

[Ramon Cota]

To summarize :
in order to get a perfect step response you have to use time aligned ( either by dsp delay or physical offset ) drivers AND first order IIR filters ? You cannot achieve perfect step response with L-R IIR filters 2 order or higher even with time aligned drivers ?
In order to get a perfect step response + flat phase you have to additionally use FIR DSP crossovers ?
Did i get this right ?

 

[gnarly]

To summarize :
in order to get a perfect step response you have to use time aligned ( either by dsp delay or physical offset ) drivers AND first order IIR filters ? You cannot achieve perfect step response with L-R IIR filters 2 order or higher even with time aligned drivers ?
In order to get a perfect step response + flat phase you have to additionally use FIR DSP crossovers ?
Did i get this right ?

I'd say pragmatically correct, with a couple of clarifications.

The first order IIR crossover needs to be an acoustic first order. An electrical first order will add to the drivers' natural rolloff and gives response that exceeds acoustic first order. Twill mess up the pretty triangle.

It is possible to use higher order IIR filters and achieve zero phase without FIR.
Subtractive delayed crossovers can do it; so can cascading a host of all-pass filters.
See posts from @gberchin for his excellent work on MDS crossovers. Run and hide from any work you find on cascading all-pass filters Lol!!

It's worth noting both of those methods introduce constant fixed latency (in the same manner that linear-phase FIR does)
Me motto: "It takes time to fix time"

I've used both of those methods to achieve zero phase and time alignment, in comparisons to linear phase xovers.
Personally, I can't find any justification to use anything other than simple FIR based linear-phase xovers.
My method of comparison is to build either MDS or cascaded all-pass as best I can, and see what latency that gives.
Then, I restrict the use of FIR to that same latency or less, and compare transfers and impulses.
I've yet to find a case where simple FIR isn't the way to go for me.

 

[gberchin]

I can't find any justification to use anything other than simple FIR based linear-phase xovers.

@gnarly This statement puzzles me a bit. Deriving a MDS crossover from a linear-phase FIR lowpass filter is actually simpler to implement than using separate FIR filters for the lowpass and the highpass. Do you implement the highpass filter separately in order to achieve a steeper transition band slope?

 

[gberchin]

In order to get a perfect step response + flat phase you have to additionally use FIR DSP crossovers ?

You can derive MDS crossovers from IIR, nonlinear-phase lowpass filters, and still achieve perfect summed step and phase responses.

 

[analogonly]

Can someone comment on "blended' crossovers--initially 1st order then transitioning to higher orders?

 

[norman bates]

I think bob brines said it was ok to do (rolling 6db for an octave then shifting to 12db an octave away from crossover point.

With digital crossovers and eq, you can make the drivers roll at 6db.

Passively crossed is is rough without zobels on the woofers. And you usually have to deal with tweeter resonance unless it is maybe 3 octaves away from crossover point, let alone the woofer or teeeter’s natural roll off.

 

[gberchin]

Passively crossed is is rough without zobels on the woofers. And you usually have to deal with tweeter resonance unless it is maybe 3 octaves away from crossover point

BTW, you can also Zobel the resonance peak: https://community.klipsch.com/topic/187697-measurement-systems/#findComment-2436802

 

[Duke]

Can someone comment on "blended' crossovers--initially 1st order then transitioning to higher orders?

I've used this approach with passive crossovers, starting out acoustically 1st or quasi-2nd order in the crossover region, then accelerating (acoustically and often electrically) to higher orders away from the crossover region. I think it offers a good combination of attributes.

 

[gnarly]

@gnarly This statement puzzles me a bit. Deriving a MDS crossover from a linear-phase FIR lowpass filter is actually simpler to implement than using separate FIR filters for the lowpass and the highpass. Do you implement the highpass filter separately in order to achieve a steeper transition band slope?

Hi Greg, yes I do like to use higher-order xovers than MDS allows (or at least allows at my skill of MDS implementation).

There's also the issue of ease of implementation. I use an open-architecture processor, which I think makes whatever method we want to use about as easy as it gets. You may remember my last foray into MDS on DIY; you gave me a good pointer how to simply my schematic. Here's it is again for the 4-way i was playing with.

Instead of all the MDS circuitry above, I can use lin-phase hpf and lpf like below.
The FIR high-pass and low pass are qsys inventory items; don't even need to generate them. Although they do entail having different freq dependent delays, hence the delay matrix. I can automate xover frequencies and delays (but in fairness I could do that for MDS as well)

Or if I want, I can just generate one filter per pass-band with both hp and low pass in, with or without pass-band acoustic target matching. Gotta love target matching FIR !
If FIR files are same size, delays not even needed. If not the same size, just insert delay into matrix.

It's sooo much easier for me than MDS, and I can use any order desired.
Hope that all made sense.